concurring and dissenting:
I concur in my brothers’ conclusion that the insurance companies were entitled to judgment in their favor. I cannot agree that the remaining defendants were enti-*493tied to judgment because the existence of triable issues of material fact foreclosed that conclusion.
The gravamen of plaintiffs’ complaint is that the defendants combined to coerce the electrical contractors of Hawaii, especially the non-members of the defendant electrical contractors’ trade association, to purchase their workmen’s compensation insurance through a single agent, defendant Oda. The combination virtually excluded plaintiffs from becoming insurance agents in the electrical contractors’ market for workmen’s compensation policies. The pleadings adequately averred a per se violation of Section 1 of the Sherman Act (15 U.S.C. § 1) unless the defendants’ program was immunized from antitrust liability either by the McCarran-Ferguson Act (15 U.S.C. §§ 1011-15) or by the so-called non-statutory labor exemption. Immunity under the former Act does not exist if the defendants used coercion in furtherance of their plan. Moreover, immunity under the non-statutory labor exemption does not exist unless Oda’s role as sole agent of record was strictly necessary for the survival of the plan. The record reveals that triable issues of fact exist both in determining the existence or non-existence of coercion and in deciding whether the designation of a sole agent was essential to the survival of the plan. Accordingly, I would reverse the district court's judgment and remand the case as to all defendants, other than the insurance carriers, for trial on those two issues.
Defendant Local 1186, International Brotherhood of Electrical Workers (IBEW), whose business manager is defendant Akito Fujikawa, represents employees of electrical contractors in the State of Hawaii. Defendant Pacific Electrical Contractors Association (PECA), whose executive secretary is defendant Walter Oda, is the trade association of the Hawaiian electrical contractors. When this claim arose, 63 of the 128 electrical contractors in Hawaii were dues-paying members of PECA. PECA negotiated with IBEW on behalf of all the electrical contractors in Hawaii. Their collective bargaining agreement covered the entire industry in the state. The remaining defendants are the Insurance Company of North America (INA), and its wholly-owned subsidiary, the Pacific Employers Insurance Company (PEIC), carriers of workmen’s compensation insurance.
Plaintiffs are general agents and solicitors of insurance in Hawaii. General agents are companies authorized by insurance carriers to solicit applications for insurance and effectuate contracts. Solicitors are salesmen appointed by general agents to solicit insurance applications and to collect premiums on a commission basis. (See Haw.Rev.Stat. §§ 431-361, 431-363.)
During the course of the renegotiation of the IBEW-PECA collective bargaining agreement in the spring of 1973, the union proposed that the contractors change the method by which they obtained workmen’s compensation insurance. Instead of having each contractor individually purchase insurance from one of the 22 carriers and through one of the 20 agencies servicing the electrical contractors’ industry, IBEW urged that insurance be provided either through a Taft-Hartley trust fund, which would have effectively made the contractors self-insurers, or from a single carrier. After receiving the report of the Martin E. Segal Company (Segal), a consulting firm, IBEW discarded the trust fund idea and recommended the adoption of a “safety group” plan. Under the plan, insurance would be purchased centrally from one carrier. Premiums would be paid individually by each contractor according to the special experience of his business, but dividends would be calculated according to the safety record of the employers as a group. In September, 1973, IBEW officials met with the PECA Board of Directors, and the PECA directors agreed to recommend that the association’s members adopt the safety group plan and mandate the participation of all employers covered by the collective bargaining agreement. In October, the membership of PECA met and approved the adoption of an amendment to Article XIX of the collective bargaining agreement establishing the safety group (“Association *494Dividend Group Plan”) and requiring participation of all electrical contractors in the plan to be administered by PECA.
PECA and IBEW developed this insurance package in response to a variety of concerns. The union apparently believed that centralized administration of insurance would lead to better claims service through increased industry leverage with a single carrier and that a single insurer would have the incentive and the opportunity to create adequate safety and rehabilitation programs. The group plan was also expected to benefit the contractors. Insurance carriers were considered more likely to vote favorable dividends because of the large volume of business at stake, and the group calculation and payment of dividends meant that small employers, who paid premiums below the threshold hitherto necessary to entitle them to dividends, would now receive dividends. Segal predicted that total dividends under the plan would be 15 percent greater than the sum of the dividends the contractors were receiving individually.
However, the welfare of the employees and the contractors were not the IBEW’s or the PECA’s only interests. One of IBEW’s purposes was to provide PECA with an additional source of income, particularly one which would enable the association to tithe non-member contractors. When faced with PECA’s initial reluctance to accede to the single-carrier proposal, IBEW argued that the association could supplement its income by charging non-members a fee for administering the plan.1 The dividends which the safety group plan would make available were soon seen as a means of supplementing PECA’s income while penalizing contractors who continued to adhere to their non-member status. According to Oda, “Blackie’s idea was to off-set dues payments by dividends and the non-members be required to forfeit their dividends since no credit can be extended towards dues contribution.”2 Dividends were not guaranteed and would not begin to be generated until a year or two after the plan was implemented. As Oda acknowledged, there was a serious “question of illegality in the allocation of dividends to dues, especially the forfeiture of dividends by non-members.”
In contrast, commissions from the insurance carrier to the agent or solicitor are paid when premiums are paid by the insured. Commissions were guaranteed regardless of the safety record of the industry and would be available from the outset of the plan. A reasonable estimate of anticipated income could be made because premiums are prepaid. Oda had been a licensed insurance solicitor since 1957, although he was inactive during the years prior to the adoption of the safety group plan. IBEW and PECA proposed that Oda be made “agent of record” for the contractors’ workmen’s compensation insurance under the plan. Using his Yamada Insurance Agency, Ltd., Oda would write and sell group plan policies. The commissions he received, after taxes, would be paid over to PECA. PECA thus would receive the income of the insurance agent operating over the entire Hawaiian electrical contractors’ industry, association members and non-members alike.
The union was solicitous of the welfare of the trade association, and it was convinced that an important aspect of the group plan was that it would provide PECA with sore*495ly-needed income. When Fujikawa first broached the subject with PECA directors, in May, 1973, he prefaced his proposal by “express[ing] the concern that the PECA is going broke.” (Minutes of May 23 meeting, supra.) IBEW’s concern for the health of PECA and its willingness to help PECA financially through the joint development of the plans and through payments by nonmember contractors predated the insurance controversy. In proposing the group plan, Oda wrote the IBEW local that the plan would “in some small measure, ‘atone’ for having promised us an Industry Promotion Fund, having it specified in the contract, but only to be overridden by the International Office [of the union]. It was intended to use the Promotion Fund concept to supplement our income as well as to assess the non-members their ‘fair share’ in managing this organization.” (October 19 Letter to Jacobson, supra.) PECA and Oda were also aware that the trade association was in precarious financial shape and that the safety group plan might relieve that condition:
“One of the incentive[s] for the PECA to undertake the program is supplemental income to off-set the declining reserves. Since the plumbers and sheet metal associations severed their affiliation with us two years ago, we have not been able to generate sufficient income to overcome the deficit.”
(Letter of Oda to Jacobson, October 8, 1973.)
Amended Article XIX of the IBEW-PECA collective bargaining agreement said nothing about choosing insurance agencies to service the group plan or about the number of such agencies there would be. But it is apparent that IBEW and PECA contemplated that the workmen’s compensation program would be set up as a “closed,” rather than an “open” plan. “A closed program is one which has a single broker of record or agent — precluding participation by any other agent or broker — even though the other agent or broker might represent the same insurance company.” (405 F.Supp. 99, 103.) Under an open plan any local agent or broker representing the carrier may place individual accounts within the program. (Id.) While it appears that all of the insurance in one workmen’s compensation safety group plan must be written by one company, there is nothing in the nature of such a program that requires that all of the insurance be written through a single insurance agent. As a PEIC vice president wrote to an official of the company’s San Francisco Service Office, “[a]ll of our Safety Groups are ‘open’ groups meaning that any properly licensed producer can place business in the group subject to our internal requirements.” (Letter of Frank C. Collins, Vice President-Underwriting, to Donald Fey, March 8, 1974.) Neither INA nor PEIC saw any reason to mandate a closed group plan to assure the success of the program. On the contrary, upon notification of the PEIC’s selection as the carrier for the electrical contractors’ group plan, an INA official wrote to Segal “to go on record indicating that the above group will be an ‘open’ group as all our safety groups are.” (Letter of Elmo Aghabeg, Manager, Workmen’s Compensation Dep’t, INA, to Jacobson, February 1, 1974.)
Despite the silence of the collective bargaining agreement, the opposition of INA/PEIC, and the lack of any structural need for a “closed” plan in the development or implementation of the safety group program, Oda responded to the financial imperatives of PECA and treated the plan as “closed.” Benefiting from the symbiotic nature of his double role as executive secretary of PECA (and hence the administrator of the safety group, as designated by the union contract) and insurance solicitor, he wrote all of the signatory contractors on January 31, 1974, advising them that the terms of the new Article XIX of the collective bargaining agreement required their participation in the safety group plan. He sent them applications for INA policies and told them to return the completed applications to him. He asked that each contractor notify him as to the expiration date of its current workmen’s compensation policy, and he sent numerous follow-up letters to the contractors who had not yet acquired *496group plan policies. Plaintiffs contend that Oda coerced contractors into purchasing INA/PEIC policies through him. Defendants characterize Oda’s activities as nothing more serious than “aggressive solicitation,” and they argue that while Oda advised contractors that the bargaining agreement required use of INA/PEIC as the carrier, Oda never told contractors that they had to buy their insurance through him. Whether or not the plan was technically “closed,” Oda and PECA actively opposed allowing other insurance agents to write INA policies under the group plan. Trenor Thompson, the president of Bayly, Martin & Fay of Hawaii, Inc. (BMF), one of the plaintiffs herein and a licensed INA agent, attested that in June, 1974, a PECA representative told him that he “would not be allowed to sell or service workmen’s compensation insurance business with any of the signatory contractors to the labor agreement ... regardless of whether such insurance was placed with [PEIC].” (Affidavit of Trenor Thompson.) When INA subsequently affirmed that BMF would be allowed to process workmen’s compensation insurance for Hawaiian electrical contractors, a Yamada official told BMF that its existing INA appointment was not enough and that an additional appointment from the PEIC was required. (Deposition of Thomas Taka-mune, President of Yamada.) According to BMF, by the time it was able to obtain that appointment, it had lost a client under the PECA plan to Oda and Yamada. Similarly, in July, 1974, Oda and PECA summarily rejected the proposal of Atlas Insurance Agency Ltd. that it also serve as agent of record under the group plan, with contractors having the right to choose freely between them. This was unacceptable, Oda wrote, because it “would defeat our secondary objective of using the earned commissions, less general excise and personal income taxes, to pay for part of our expenses.”
Although a few agents other than Ya-mada/Oda did manage to sell INA/PEIC workmen’s compensation insurance policies to electrical contractors, Oda’s success in implementing a safety group plan effectively closed to all agents other than himself is manifest: Between February 1, 1974, and June 20, 1975, all but 14 of the 128 contractors bound by the IBEW-PECA collective bargaining agreement placed their INA/PEIC policies through Oda’s Yamada agency.
I
If the plaintiff’s antitrust attack were based upon a challenge to the single-carrier aspect of the safety group plan, I would share both the district court’s and my colleagues’ reluctance to endorse plaintiffs’ assault on the plan. The plan has obvious benefits for both contractors and employees, open bidding was used for the selection of the single carrier, and several hundred such programs presently exist in the United States. Plaintiffs’ attack, however, is not on the single carrier feature of the plan, but on defendants’ alleged agreement to use Oda as the single agent.
Plaintiffs’ contention is that the defendants combined to make Oda the sole insurance agent for all of the electrical contractors in Hawaii for the purpose of providing PECA with income from commissions generated by the sale of the policies to all of the contractors, PECA members and nonmembers alike. They also claim that the defendants effectuated their “closed” plan by coercive means, such as threatening nonmember contractors with IBEW reprisals if the non-members failed to authorize Oda to process their INA/PEIC applications. The result was, according to the plaintiffs, that they were denied access to the non-PECA contractors’ market for INA/PEIC workmen’s compensation policies.
Market foreclosure and coercion are the classic elements of a violation of Section 1 of the Sherman Act; collaborative conduct characterized by these factors is per se unlawful under the Act. (See, e. g., United States v. General Motors Corp. (1966) 384 U.S. 127, 86 S.Ct. 1321, 13 L.Ed.2d 415; Radiant Burners, Inc. v. Peoples Gas Light & Coke Co. (1961) 364 U.S. 656, 81 S.Ct. 365, 5 L.Ed.2d 358; Klor’s Inc. v. Broadway-*497Hale Stores, Inc. (1959) 359 U.S. 207, 79 S.Ct. 705, 3 L.Ed.2d 741; Helix Milling Co. v. Terminal Flour Mills Co. (9th Cir. 1975) 523 F.2d 1317, 1320-21; E. A. McQuade Tours, Inc. v. Consolidated Air Tour Manual Committee (5th Cir. 1972) 467 F.2d 178, 186-87; 11 J. Von Kalinowski, Antitrust Laws and Trade Regulation 76.02, at 76-11 to 76-12 (1977); Barber, Refusals to Deal Under the Federal Antitrust Laws, 103 U.Pa.L.Rev. 847, 872-79 (1955).)
The collaborative conduct attacked does not fit snugly within any of the more familiar kinds of group boycotts. (405 F.Supp. 99, 107-08.) The combination being challenged is unusual and the imposition by a trade association and a union of a “closed” agency plan in the context of an insurance safety group program is novel. But the Sherman Act is “couched in broad terms . adaptable to the changing types of commercial production and distribution that have evolved since its passage.” (United States v. E. I. duPont de Nemours & Co. (1956) 351 U.S. 377, 386, 76 S.Ct. 994, 100 L.Ed. 1264.) No “all-fours” resemblance to arrangements previously held unlawful is necessary when the challenged combination is marked by the indicia of per se illegality:
“The distinguishing feature of the group boycott cases is group action to coerce third parties to conform to the pattern of conduct desired by the group. . Such action offends the concept of a free market because it places involuntary restraints on the trading opportunities of strangers to the group.”
(Barber, supra, 103 U.Pa.L.Rev. at 875.) In this case the “group” is composed of the defendants; the third parties coerced are the non-member contractors who, plaintiffs contend, were compelled to buy their INA/PEIC policies from Oda’s Yamada agency; and the “strangers” whose trading opportunities have been restrained by defendants’ activities are the plaintiff insurance agents who were denied access to the non-member contractors’ market for workmen’s compensation policies. As the Fifth Circuit has observed, “the touchstone of per se illegality has been the purpose and effect of the arrangement in question. Where exclusionary or coercive conduct has been present, the arrangements have been viewed as ‘naked restraints of trade,’ and have fallen victims to the per se rule.” (E. A. McQuade Tours, Inc. v. Consolidated Air Tour Manual Committee, supra, 467 F.2d at 187.) Plaintiffs contend that both exclusionary and coercive conduct on the part of the defendants are present. If they are right, per se illegality follows.
Although the interrelationships of the defendants and the context are unusual, the defendants’ arrangement nevertheless resembles combinations that are characteristic of vertical boycotts. The district court found that there was no “vertical combination among IN A at one level, and IBEW/PECA at the other, to exclude from the market one of their competitors.” (405 F.Supp. at 108.) The district court apparently assumed that PECA was merely a trade association of electrical contractors which was not in competition with general agents in the insurance field. However, under the circumstances of this case, PECA was not just a trade association, nor was Oda simply a trade association’s executive secretary. Within the structure of the safety group, Oda was an insurance solicitor selling INA/PEIC workmen’s compensation policies to 114 of the 128 electrical contractors. PECA, as recipient of Oda’s commissions, in a very real sense was in the insurance agency business. IBEW had a stake in PECA’s fiscal welfare, and it was at least an incidental beneficiary of Oda’s insurance sales.
PECA was not merely the passive beneficiary of the fortuitous circumstance that its executive secretary possessed an insurance solicitor’s license and a generous disposition toward his employer which led him to give it his commissions. Rather, the evidence demonstrates that PECA and IBEW implemented the plan with the idea that Oda would act as an insurance agent and turn over his commissions to the association; that PECA, through Oda, could use its position as administrator of the plan to maxim*498ize its insurance commission income by telling contractors (especially non-members of the association) to take out INA/PEIC policies through Oda using the leverage of the collective bargaining agreement. Oda/PECA were in both the electrical contractors’ field and in plaintiffs’ insurance solicitation business. This curious double role was used to promote their competitive position in the safety group insurance sales’ market.
Once it is recognized that, through Oda, PECA was in the insurance sales business, the resemblance of this arrangement to the classic vertical boycott model emerges. In Klor’s Inc. v. Broadway-Hale Stores, Inc., supra, 359 U.S. 207, 79 S.Ct. 705, 3 L.Ed.2d 741, plaintiff Klor’s, a small retail appliance store, charged that Broadway-Hale, a department store chain, had used its “ ‘monopolistic’ buying power” to obtain the agreement of the major manufacturers and distributors of appliances not to sell to Klor’s, or to sell to it only at discriminatory prices. This foreclosure of one business by another from a market they both sought was held to be per se unlawful, because both the business excluded and the other businesses would no longer have the opportunity to deal with it:
“Alleged in this complaint is a wide combination consisting of manufacturers, distributors and a retailer. . This combination takes from Klor’s its -freedom to buy appliances in an open competitive market and drives it out of business as a dealer in the defendants’ products. ‘ It deprives the manufacturers and distributors of their freedom to sell to Klor’s at the same prices and conditions made available to Broadway-Hale, and in some instances forbids them from selling to it on any terms whatsoever.” (359 U.S. at 212-13, 79 S.Ct. at 710.)
Klor’s and Broadway-Hale were in competition for the purchase of appliances. Here, PECA, through Oda and with the support of IBEW, is in competition with plaintiffs in the sale of INA/PEIC workmen’s compensation policies to members of the safety group. Defendants’ combination has effectively driven plaintiffs out of - business as insurance agents with respect to the contractors’ workmen’s compensation policies, and it has deprived the contractors — especially those who are not members of PECA and therefore lack any stake in providing the association with additional sources of income — of their freedom to buy their INA/PEIC policies from the agents of their choice.
If proved, the twin evils of coercion and market foreclosure would together be sufficient to render defendants’ arrangement per se unlawful. The district court appeared to assume that there was an implicit agreement to use Oda as the exclusive agent under the group plan (405 F.Supp. at 109), but it held that no per se violation existed because the impact on plaintiffs was unintentional. It was the incidental result of the otherwise salutary group plan, and not the product of an anticompetitive motive. Applying Joseph E. Seagram & Sons, Inc. v. Hawaiian Oke & Liquors, Ltd. (9th Cir. 1969) 416 F.2d 71 (agreements having the effect of excluding a party from a market are not per se unlawful where the exclusion was a “merely incidental result” of actions aimed at attaining a legitimate business objective), the district court found that the “closed” agency arrangement presented no per se violation and proceeded to test it according to the “rule of reason.”
By its very terms, however, Hawaiian Oke is inapposite. In Hawaiian Oke, two liquor manufacturers agreed that in order to obtain improved distribution services, they would transfer their exclusive distributorships from the plaintiff to another distributor. We held that that agreement was not a per se violation of the Sherman Act. The holding in Hawaiian Oke was contingent on the finding that defendants’ agreement had no “adverse purpose or effect on competition.” (416 F.2d at 78.) A combination which is anticompetitive in purpose or effect, however, does constitute an unreasonable restraint of trade in violation of the Sherman Act. (See, e. g., Fount-Wip, Inc. v. Reddi-Wip, Inc. (9th Cir. Feb. 6, 1978) 568 F.2d 1296, 1300-1301; Mutual Fund Investors, Inc. v. Putnam Manage-*499meat Co. (9th Cir. 1977) 553 F.2d 620, 626; Bushie v. Stenocord (9th Cir. 1972) 460 F.2d 116, 119.)
Plaintiffs here contend that defendants’ combination was anticompetitive both in effect and in intent. The exclusionary effects of defendants’ activities are clear: only 14 of the 128 electrical contractors in Hawaii bought their INA/PEIC insurance from an agent other than Oda. Bearing in mind the Supreme Court’s admonition that “[e]limination, by joint collaborative action, . from access to the market is a per se violation of the [Sherman] Act” (United States v. General Motors Corp. (1966) 384 U.S. 127, 145, 86 S.Ct. 1321, 1330, 13 L.Ed.2d 415), we have been particularly sensitive to the antitrust implications of agreements which have resulted in the virtual exclusion from the relevant market of those businesses which are not parties to the challenged combination. (See, e. g., Helix Milling Co. v. Terminal Flour Mills Co., supra, 523 F.2d at 1320-22.)
The exclusionary effect of defendants’ combination was not the incidental consequence of actions designed to attain the legitimate business objective of the safety group. As INA has reiterated in its communications to Segal, Oda and PECA, a closed agency arrangement is not a prerequisite for the success of a safety group, most safety groups are in fact “open,” and INA apparently opposed Oda’s efforts to impose a closed situation. Because an exclusive agency was not normally a part of a safety group plan, it cannot be concluded at this summary state that the exclusive agency was an unintentional aspect of the safety group, or an incidental consequence of the other arrangements. The novelty of the closed agency aspect of this arrangement suggests that exclusion was intended. Moreover, the communications among IBEW, PECA, and Segal indicate that PECA’s opportunity of reaping commission income was a definite consideration in the minds of the principal actors — a reinforcement of the inference that the closed aspect of the insurance solicitation was intentional.
An intent to exclude is “anticompetitive.” Oda/PECA may not have intended to drive the plaintiffs out of this business, and they may have had no animosity specifically directed against any of the plaintiffs. But, within the limited context of the sale of workmen’s compensation insurance to electrical contractors, plaintiffs and Oda/PECA were in competition. To the extent that defendants’ combination was designed to enable Oda/PECA to capture the contractors’ market in INA/PEIC policies, it was equally aimed at excluding the plaintiffs from that market. “It is . not always necessary to find a specific intent to restrain trade ... in order to find that the antitrust laws have been violated. It is sufficient that a restraint of trade . results as the consequence of a defendant’s conduct or business arrangements.” (United States v. Griffith (1948) 334 U.S. 100, 105, 68 S.Ct. 941, 944, 92 L.Ed. 1236.) As we recently explained, proof of anticompetitive motivation does not require a showing that defendants intended to harm a competitor. (United States v. Hilton Hotels Corp. (9th Cir. 1972) 467 F.2d 1000, 1002-03.)
In Hilton Hotels, the hotels, restaurants, and hotel and restaurant supply companies of Portland, Oregon, had organized an association to attract conventions to the city. To finance the operations of the association, they required members to make certain contributions and, to enforce that requirement, the hotels agreed to curtail purchases from those suppliers who failed to make their contributions. Although we concluded that the sole purpose of defendants’ refusal to deal was “ ‘to bring convention dollars into Portland’ ...[,] the necessary and direct consequence of defendants’ scheme was to deprive uncooperative suppliers of the opportunity to sell to defendant hotels . . and to deprive defendant hotels of the opportunity to buy supplies from such suppliers Defendants therefore ‘intended’ to impose these restraints upon competition in the only sense relevant here.” (Id. at 1002.) (emphasis added)
*500Although the purpose of defendants’ combination may have been only to provide PECA with additional financial support to help it perform its various functions on behalf of the electrical contractors’ industry, the “necessary and direct consequence” of their action was to restrain the trading opportunities of both the contractors and the plaintiff insurance agents. Thus, the anticompetitive effects of the closed agency relationship were clearly “intended.” (See also Albrecht v. Herald Co. (1968) 390 U.S. 145, 149-50, 88 S.Ct. 869, 19 L.Ed.2d 998 (plaintiff independent newspaper carrier sued publisher who terminated his distributorship when plaintiff sold newspapers at a price higher than that set by the publisher; publisher hired a solicitation agency to induce newspaper readers to transfer their allegiance to another carrier; and publisher gave customers solicited by the agency to a second carrier; plaintiff’s antitrust action was upheld against the combination of publisher, solicitation agency, and second carrier although the agency’s only purpose “was undoubtedly to earn its fee”); cf. Fashion Originators’ Guild of America, Inc. v. FTC (1941) 312 U.S. 457, 467-68, 61 S.Ct. 703, 85 L.Ed. 949 (horizontal combination organized to combat style piracy in fashion industry “intended” the anticompetitive effect of boycotting retailers who dealt with style pirates).)
The presence of anticompetitive intent and effects would alone be sufficient to distinguish this case from Hawaiian Oke and to constitute a prima facie case of an unlawful restraint of trade. Plaintiffs’ assertion that defendants effectuated this restraint through coercion renders Hawaiian Oke inapposite, and, if coercion were proved at trial, would require treating defendants’ arrangement as unreasonable per se. In Hawaiian Oke, defendants’ agreement to switch distributorships was voluntary. Plaintiff presented no evidence that coercion played any role in the parties’ dealings with each other. (416 F.2d at 78. Accord Bridge Corp. of America v. American Contract Bridge League, Inc. (9th Cir. 1970) 428 F.2d 1365, 1369-70 (Hawaiian Oke applicable because plaintiffs did not argue that defendants “combined ... for the . purpose of coercing . . .”).) Plaintiffs here contend that PECA was in direct competition for the insurance sales of the contractors belonging to the safety group. The gravamen of plaintiffs’ claim is that the choice of Oda as insurance agent was imposed by the IBEW/PECA combination, and it was not the result of any voluntary choice among the contractors. The coercion element transforms defendants’ combination from an arguably unreasonable one into one which is unlawful per se.
The pivotal question is whether the record before the district court showed that the coercion issue presented no triable issue of fact. Of course, plaintiffs have the burden of proving coercion because that element is essential both to defeat the defendants’ immunity arguments and to support the plaintiffs’ claim of a per se violation. On summary judgment, however, the burden was upon the defendants, who moved for summary judgment, clearly to establish the non-existence of any genuine issue of fact material to judgment in their favor. (E. g., SEC v. Koracorp Industries, Inc. (9th Cir. 1978) 575 F.2d 692, 697; 6 Pt. 2 J. Moore, Federal Practice ¶ 56.15[4], p. 56-511; id. U 56.15[8], p. 56-643.) Defendants, other than the insurance carriers, failed to carry that burden. The burden is especially difficult to meet when the material fact asserted to be in dispute is coercion, because proof of that issue turns on such subjective factors as defendants’ motives and intent, the credibility of their assertions, and contractors’ reasonable perceptions of defendants’ conduct. (Poller v. Columbia Broadcasting System, Inc. (1962) 368 U.S. 464, 473, 82 S.Ct. 486, 7 L.Ed.2d 458.) Plaintiffs have articulated the basis of their claim for violation of the Sherman Act and they have succeeded in producing enough evidence to raise a triable issue with respect to coercion.
II
I next turn to the question whether the defendants were immunized from antitrust liability under the McCarran-Ferguson Act *501(15 U.S.C. §§ 1011-15). The Act was passed in response to United States v. South-Eastern Underwriters Association (1944) 322 U.S. 533, 64 S.Ct. 1162, 88 L.Ed. 1440. Before South-Eastern Underwriters, issuing a policy of insurance was not thought to be a “transaction of commerce.” (Paul v. Virginia (1869) 75 U.S. (8 Wall.) 168, 183, 19 L.Ed. 357.) When the Supreme Court decided that insurance transactions were subject to federal regulation under the commerce clause and that the antitrust laws were applicable to them, Congress enacted the McCarran-Ferguson Act. The Act states that “the continued regulation and taxation by the several States of the business of insurance is in the public interest.” (15 U.S.C. § 1011.) “Obviously Congress’ purpose was broadly to give support to the existing and future state systems for regulating and taxing the business of insurance.” (Prudential Insurance Company v. Benjamin (1946) 328 U.S. 408, 429, 66 S.Ct. 1142, 90 L.Ed. 1342.)
The McCarran-Ferguson Act’s exemption of the insurance business is fitful. The Act provides that the antitrust laws “shall be applicable to the business of insurance to the extent that such business is not regulated by State law.” (15 U.S.C. § 1012(b).) The Act also provides that the Sherman Act would remain applicable to “any agreement to boycott, coerce, or intimidate, or act of boycott, coercion, or intimidation.” (15 U.S.C. § 1013(b).) We need not dwell upon whether the acts alleged to violate the antitrust laws were part of the “business of insurance,” or whether the business of insurance was “regulated by State law,” because the plaintiffs have alleged sufficient facts to bring them within the coercion exclusion from the exemption. (E. g., Ballard v. Blue Shield of Southern West Virginia, Inc. (4th Cir. 1976) 543 F.2d 1075, 1078.)
IBEW and Fujikawa also claim that they are immune from antitrust liability under the so-called labor exemption. The sources of the labor exemption from the antitrust laws are those provisions of the Clayton Act (15 U.S.C. § 17 and 29 U.S.C. § 52) and the Norris-LaGuardia Act (29 U.S.C. §§ 104, 105, 113) which declare that labor unions are not combinations or conspiracies in restraint of trade and which specifically exempt certain union activities, such as secondary picketing and group boycotts, from the reach of the antitrust laws. (Connell Construction Co. v. Plumbers & Steamfitters Local Union No. 100 (1975) 421 U.S. 616, 621-22, 95 S.Ct. 1830, 44 L.Ed.2d 418.)
The statutory exemption was designed to insulate legitimate collective activity by employees, which is inherently anticompetitive but favored by national labor policy, from the proscriptions of the antitrust laws. (See Apex Hosiery Co. v. Leader (1940) 310 U.S. 469, 60 S.Ct. 982, 84 L.Ed. 1311.) As Mr. Justice Frankfurter explained in United States v. Hutcheson (1941) 312 U.S. 219, 232, 61 S.Ct. 463, 466, 85 L.Ed. 788, the statutory exemption immunizes from antitrust liability all legitimate labor activities undertaken by a union in furtherance of its own interests “[s]o long as [the] union . does not combine with non-labor groups.” Where concerted action or agreements between unions and “non-labor groups” is involved, the statutory exemption is unavailable. (Allen Bradley Co. v. Local Union No. 3, IBEW (1945) 325 U.S. 797, 65 S.Ct. 1533, 89 L.Ed. 1939.) In order properly to accommodate the congressional policy favoring the peaceful resolution of employer-union disputes through bargaining, the Supreme Court has held that certain union-employer agreements must be accorded a limited non-statutory exemption from antitrust sanctions. (E. g., Connell Construction, supra.)3
*502The contours of the non-statutory labor exemption have been developed by the Supreme Court in four major decisions over the past thirty years. Although the Court’s opinions were tailored largely to the fact situations of the particular cases with which it was confronted, a few general principles useful in determining the availability of the labor exemption have emerged. First, if a union-employer agreement has a direct and immediate impact on competition in some product market then, subject to the principles discussed further below, the parties to the agreement may be liable under the antitrust laws without the shield of the labor exemption. (See Connell Construction, supra, 421 U.S. at 624-25, 95 S.Ct. 1830; Local Union No. 189, Amalgamated Meat Cutters & Butcher Workmen v. Jewel Tea Co. (1965) 381 U.S. 676, 691, 85 S.Ct. 1596, 14 L.Ed.2d 640; UMW v. Pennington (1965) 381 U.S. 657, 663, 85 S.Ct. 1585, 14 L.Ed.2d 626; Allen Bradley, supra, 325 U.S. at 808-11, 65 S.Ct. 1533.) Agreements which directly fix prices or deny access to the business market to competitors of the employers who are parties to the agreement are particularly suspect, and the labor exemption is very unlikely to be available in such a situation. (See Pennington, supra, 381 U.S. at 663, 85 S.Ct. 1585 (a union-employer agreement designed to protect wages through fixing prices cannot be shielded from antitrust liability by the labor exemption); Allen Bradley, supra, 325 U.S. at 809, 65 S.Ct. 1533 (a union may not seek to increase job opportunities for its members by combining with employers to eliminate the employers’ competitors from the market); Bodine Produce, Inc. v. United Farm Workers Organizing Comm. (9th Cir. 1974) 494 F.2d 541, 551 (“union cooperation which enables one or more employers to obtain control of the supply and price of a certain product in a particular market, or to make possible the elimination of troublesome competition, is unmistakably tainted”); P. Areeda, Antitrust Analysis (2d ed. 1974) at 104 (“there is no exemption where the union agrees with one or more employers to deny competing employers access to the market”).)
Second, a union may not combine with one set of employers and agree to impose terms of a union-employer accord on other employers who are competitors of the first employers and not parties to the agreement. This principle emerged in Pennington where the complaint charged that the UMW had conspired with the large coal operators to eliminate the smaller companies by agreeing to impose the terms of the 1950 Wage Agreement existing between the large operators and the union on all the coal companies regardless of a particular company’s ability to pay. Although the Court found that the agreement concerned wages “the very heart of those subjects about which employers and unions must bargain” and that the effect on the product market “results from the elimination of competition based on wages which is not the kind of restraint Congress intended the Sherman Act to proscribe” (381 U.S. at 664, 85 S.Ct. at 1590), the agreement to impose the wage terms on the small coal operators was held subject to the antitrust laws because it operated directly on employers who were not parties to the agreement.4
*503Third, where an agreement has an immediate impact on the product market, federal labor policy is implicated sufficiently to prevail and to protect the agreement from antitrust liability only where the agreement directly concerns a mandatory subject of bargaining. (See Jewel Tea, supra, 381 U.S. at 680, 85 S.Ct. 1596; Pennington, supra, 381 U.S. at 665, 85 S.Ct. 1585; Mack-ey, supra, 543 F.2d at 614.) But the fact that a restriction on competition is “intimately related” to the attainment of a legitimate labor goal, even if it is a mandatory subject of bargaining, while a necessary prerequisite to the labor exemption is not a sufficient condition. (Pennington, supra, 381 U.S. at 664-65, 85 S.Ct. at 1590 (“This is not to say that an agreement resulting from union-employer negotiations is automatically exempt from Sherman Act scrutiny simply because the negotiations involve a compulsory subject of bargaining . . .”). See also Jewel Tea, supra, 381 U.S. at 694, 85 S.Ct. 1596; but cf. Jewel Tea, id. at 732-33, 85 S.Ct. 1596 (Goldberg, J., concurring, would exempt all agreements directly concerning mandatory subjects of bargaining).)
Where an agreement has both a direct and substantial impact on competition among businessmen and an immediate and concrete effect on a mandatory subject of bargaining, the Court has first determined whether the agreement’s anticompetitive effect is an incidental, derivative consequence of the effort to effectuate labor’s legitimate objective, or, rather, whether the anticompetitive effect is independent of the restriction directly necessary to serve labor’s concerns. The application of this “least restrictive means” approach is illustrated in Jewel Tea. There, a Chicago retail meat market chain sued seven union locals and the Chicago area association of retail food stores, alleging that the defendants had conspired to restrain competition among the retail meat markets by limiting the marketing hours for the sale of fresh meat, through a clause in the collective bargaining agreement between the association and the union and in the agreement between the plaintiff and the unions. Jewel Tea contended that the purpose of the marketing hours restriction was to aid small butcher shops by preventing self-service meat markets, like plaintiff’s, from operating at night when the smaller shops were closed. The Court decided that the effect of the restriction on competition in the meat marketing business was “apparent and real.” (381 U.S. at 691, 85 S.Ct. 1596.) Unlike Pennington, the Court found there was no claim that the union and small shop owners had conspired together to impose the marketing restrictions on Jewel Tea. (Id. at 688, 85 S.Ct. 1585.) Turning to the labor aspect of the agreement, the Court concluded that the specific hours of employment, as well as the number of hours, were a component of the working hours which had been considered a mandatory subject of bargaining. The marketing hours restriction thus operated in an area where “the concern of union members is immediate and direct.” (Id. at 691, 85 S.Ct. at 1603.) The question for the Court then became one of fact: Were night operations feasible without butchers and without infringing on butchers’ interests? If so, then the agreement was overbroad because it went beyond the direct implementation of butchers’ concerns and unnecessarily restricted competition. If not, the anti-competitive consequence was simply a derivative, incidental result of the union’s legitimate effort to fix the hours of work. Adopting the district court’s conclusion that night operations without butchers were infeasible, the Court held that the labor exemption was available to shield the marketing hours clause from antitrust liability. (Id. at 694-97, 85 S.Ct. 1585.) (See also Connell Construction, supra, 421 U.S. at 624, 95 S.Ct. at 1837 (agreement had “significant adverse effects on *504the market and on consumers — effects unrelated to the union’s legitimate goals of organizing workers and standardizing working conditions.”).)
IBEW’s alleged agreement with the insurance carriers and with PECA, together with PECA’s arrangement with Oda, had a direct and immediate effect of excluding plaintiffs from the relevant market. This arrangement operated to deny plaintiffs, who were Oda/PECA’s competitors, access to the non-PECA contractors. That agreement is “unmistakably tainted.” (Bodine Produce, Inc. v. United Farm Workers Organizing Comm., supra, 494 F.2d at 551. Accord Allen Bradley Co. v. Local Union No. 3, IBEW, supra, 325 U.S. 797, 65 S.Ct. 1533, 89 L.Ed. 1939.) The restraint on trade primarily affected non-parties to the agreement. Plaintiffs contend that the arrangement was nothing more than an effort by IBEW and PECA to coerce the non-PECA contractors into purchasing their insurance through Oda. Although one aspect of this arrangement was that Oda/PECA were in competition with plaintiffs for insurance business within the context of the safety group, PECA contractors were normally in competition with non-PECA contractors in respect of every other phase of their contracting businesses. The closed agency agreement directly benefited PECA members by providing the association with additional income and by making possible a reduction in PECA dues. No similar direct benefits resulted to non-members. Therefore, one consequence of the agreement was to make PECA membership more attractive. Although the competitive advantage occurring to PECA members from a reduction in the dues obligation might be small, the closed agency arrangement did tend to strengthen the position of PECA and of PECA members, thereby potentially restraining competition in the electrical contractors’ field as well as in the insurance agents’ business. Thus, here, as in Pennington, the inference can be drawn that the arrangement was an effort by a union and one group of employers to impose a requirement on the employers’ competitors.
It is well established that the benefits of an insurance program are among the “non-wage” conditions of employment which have been considered mandatory subjects of bargaining. (See Allied Chemical & Alkali Workers, Local Union No. 1 v. Pittsburgh Plate Glass Co., Chemical Div. (1971) 404 U.S. 157, 159, 92 S.Ct. 383, 30 L.Ed.2d 341; Inland Steel Co. v. NLRB (7th Cir. 1948) 170 F.2d 247, 251 (accepting the Board’s conclusion “that the term ‘wages’ . must be construed to include emoluments of value, like pension and insurance benefits, which may accrue to employees out of their employment relationship”).) Whether the selection of the insurance agent for a workmen’s compensation program is also a mandatory subject of bargaining turns on whether the benefits available under the insurance scheme are inextricably bound up with the identity of the insurance agent. (Cf. Oil, Chemical & Atomic Workers v. NLRB (1976) 178 U.S.App.D.C. 278, 547 F.2d 575, 582 n. 6; Connecticut Light & Power Co. v. NLRB (2d Cir. 1973) 476 F.2d 1079, 1082-83 (only where identity of insurance carrier is closely bound up with benefits available under insurance plan is the selection of a carrier a mandatory subject of bargaining); Bastian-Blessing, Div. of Golconda Corp. v. NLRB (6th Cir. 1973) 474 F.2d 49, 54 (“We have sought to find a way to separate the carrier from the benefits in this case, and we have failed. Under these facts, the Board’s remedy that the benefits be restored by restoring ‘the preexisting . . . plan’ as a single ‘ball of wax’- appears justified.”).)
We are then left with the question whether Oda’s role, as a conduit through which commissions flowed to PECA was essential to the safety group program. If that question is answered affirmatively, then the PECA/IBEW agreement has a “direct and concrete" impact on a labor subject, and its anticompetitive effects are no more than the incidental, derivative consequences of the “labor” aspect of the agreement. If the question is answered negatively, then the closed agency arrangement has at best an indirect effect on at*505taining legitimate labor goals and its harmful impact on competition is far greater and more immediate than is necessary to serve labor interests.
In my view, the record before us is inadequate to provide a definitive answer to that critical question. The district court did not expressly address the issue. Although the court said that the labor exemption provided no protection for the IBEW/PECA plan (405 F.Supp. at 113), the court earlier found it “clear from plaintiffs’ evidence that a single broker or solicitor must coordinate the group plan if it is to achieve all of the objectives previously cited,” and further that it was “laudatory” that the coordinating broker should be willing to use his profits to cover the expenses of the program. But the only evidence that the court cited in support of its conclusion that a single broker was needed was a statement of a vice president of one of the insurance agent plaintiffs explaining that group plans may be set up on either an open or a closed basis. (See 405 F.Supp. at 111, nn. 50 & 51 and at 103, n. 11.) The only other evidence in the record is the repeated assertion by INA that an exclusive agent of record is not a necessary prerequisite to the success of a group plan. On remand, I would have the district court consider whether Oda’s role is “so intimately related” (Jewel Tea, supra, 381 U.S. at 689, 85 S.Ct. 1596) to the administration of plan as to render it necessary and further to consider whether the anti-competitive consequences of the agreement in two product markets are justified on the ground that Oda’s role was strictly necessary to the survival of a plan beneficial to the union members.
Ill
I concur with the majority’s conclusion that the insurance carrier defendants were properly granted summary judgment. INA/PEIC did not participate in any of the discussions in 1973 among IBEW, PECA, and Segal, concerning the desirability of a safety group plan, the various methods of establishing the program, and the possible benefits which the scheme might have in improving PECA’s deteriorating financial condition. The insurance carriers did not enter the picture until after the end of the year when INA/PEIC submitted its bid to be carrier for the safety group. INA played no role whatever in giving Oda an exclusive agency relationship. Indeed, it has sold safety group policies through other agencies when those other agents were available. The company has never indicated any interest in providing Oda with an exclusive designation, and it had nothing to gain from such an arrangement. Indeed, it may have stood to lose goodwill of other Hawaiian insurance agents, who sold other kinds of INA policies. From the time INA received notification that it would be the carrier for the safety group, it has repeatedly insisted that the group be an open one. It communicated this thought to Se-gal, to Oda, to the PEIC San Francisco office, to Yamada, and to PECA’s attorney. (405 F.Supp. at 106.) In short, there is no evidence in the record suggesting that INA/PEIC were in any way active in a scheme to make Oda the exclusive agent for the safety group or that they took any other steps to further the scheme.
The union and trade association defendants present a very different picture. As to them, there is substantial evidence that they combined with the common goal of utilizing the safety group scheme as means of providing the association with new income. They jointly decided to make Oda the agent of record in order to further that purpose. The union denied that it played any role in inducing contractors to purchase their insurance through Oda or that it threatened any contractor with labor trouble, or the institution of grievance proceedings, if any particular firm failed to place insurance through Oda. However, plaintiffs need not prove that the union acted coercively to implement the PECA/IBEW plan. “Proof of participation in a course of conduct that has the necessary consequence of barring entry of competition into the market would provide the basis for a finding of a combination.” (Helix Milling Co. v. Terminal Flour Mills Co., supra, 523 F.2d at 1322, citing Albrecht v. Herald Co. (1968) *506390 U.S. 145, 149-50, 88 S.Ct. 869, 19 L.Ed.2d 998.) The district court correctly decided that plaintiffs have shown that the union and the trade association participated in a common plan to use Oda’s position to enable PECA to derive income from the project.
For purpose of resisting summary judgment, the record contains enough evidence, together with inferences necessarily drawn in favor of the plaintiffs, to show a triable issue of fact in respect of claimed coercion. To be sure, Oda’s dealings with non-PECA contractors who were reluctant to buy insurance through Oda were not explicitly coercive. But when these dealings are viewed in the context of the contractors’ relationship to him and to PECA, an inference arises that coercion was implicit. In his role as executive secretary of the trade association which normally bargained with the union on behalf of the contractors, he appeared to be a disinterested person advising the contractors who were confused about their obligations with respect to workmen’s compensation. Oda answered those questions through a steady stream of letters and telephone calls in which he told the contractors that the union contact required them to take INA/PEIC policies, and he warned them that the failure to do so left them “in violation of your union contract.” He sent them INA applications and repeated follow-up letters, and directed each of them to return the application to him. The inference arises that Oda, at some point along the line, dropped his role as disinterested advisor and became the highly interested insurance solicitor. The duality of his role was matched by PECA. PECA changed from a trade association representing the interests of all the contractors in their dealings with the union into a membership organization seeking financial support from the business of non-members. The course of dealing both by Oda and by PECA was deceptive, and the inference is certainly permissible, if not absolutely compelling, that both the intent and the effect were coercive.
I would reverse the grant of summary judgment in favor of IBEW, PECA, Oda, and Fujikawa and remand the case for trial against those defendants.
. See Minutes of PECA Board of Directors meeting of May 23, 1973: “Mr. Fujikawa [the IBEW Business Manager] feels that by charging the electrical contractors a fee for this service we could supplement our income. Furthermore, non-member contractors will be required to participate through the industry plan.” See also Minutes of PECA directors meeting of September 27, 1973: “Since the non-association members would also be participating, this could be a means of assessing them for the services we provide.” Letter of Oda to Berton Jacobson, Executive Vice President of Segal, October 19, 1973: “Blackie Fujikawa has stated that the Association Dividend Group Plan will be a means for the PECA to assess the non-members for the services we have been providing them throughout the years in negotiating and servicing their union agreement.”
. Letter of Oda to Jacobson, supra. See also Minutes of meeting of September 27, supra, “the dividends received could be allocated towards off-setting dues.”
. The labor exemption applies to the union-employer agreement not just to the union. Those non-labor groups who were parties to the agreement on which antitrust liability is predicated may also avail themselves of the labor exemption shield (See Mackey v. National Football League (8th Cir. 1976) 543 F.2d 606, 612; Scooper Dooper, Inc. v. Kraftco Corp. (3d Cir. 1974) 494 F.2d 840, 847 n. 14; cf. Local Union No. 189, Amalgamated Meat Cutters & Butcher Workmen v. Jewel Tea Co. (1965) 381 U.S. 676, 729-30, 85 S.Ct. 1596, 14 L.Ed.2d 640 (Goldberg, J., concurring) (a finding that the labor exemption is unavailable to the union necessarily exposes the employer to antitrust liability). Thus, should the labor exemption be *502found applicable to immunize defendants’ agreement, all the defendants, not just 1BEW and Fujikawa, could rely on it.
Plaintiffs contend that defendants cannot raise the labor exemption on this appeal because the district court ruled against the defendants on this point below and the defendants failed to take a cross-appeal. But the failure to cross-appeal creates no bar to our consideration of the labor exemption question. “The prevailing party may, of course, assert in a reviewing court any ground in support of his judgment, whether or not that ground was relied upon or even considered by the trial court.” (Dandridge v. Williams (1970) 397 U.S. 471, 475 n. 6, 90 S.Ct. 1153, 1156, 25 L.Ed.2d 491. See also United States v. American Railway Express Co. (1924) 265 U.S. 425, 435, 44 S.Ct. 560, 564, 68 L.Ed. 1087. (“. . . the appellee may, without taking a cross-appeal, urge in support of a decree any matter appearing in the record, although his argument may involve an attack upon the reasoning of the lower court . . . .”).)
. The Eighth Circuit has generalized the rule in Pennington into a principle requiring that a restraint on trade which results from a union-*503employer agreement must “primarily [affect] only the parties to the collective bargaining relationship.” (Mackey v. National Football League (8th Cir. 1976) 543 F.2d 606, 614. See also Connell Construction, supra, 421 U.S. at 624-26, 95 S.Ct. 1830 (union-employer agreement had significant affect on employer who had no collective bargaining relationship with the union and had no employees that the union sought to represent).)